Rare earth permanent magnets prepared by hot deformation process
Chen Ren-Jie1, †, Wang Ze-Xuan1, Tang Xu1, Yin Wen-Zong1, Jin Chao-Xiang1, Ju Jin-Yun1, Lee Don2, Yan A-Ru1, ‡
CAS Key Laboratory of Magnetic Materials and Devices, Ningbo Institute of Material Technology and Engineering, Chinese Academy of Sciences, Ningbo 315201, China
University of Dayton, Dayton, OH, USA

 

† Corresponding author. E-mail: chenrj@nimte.ac.cn aruyan@nimte.ac.cn

Abstract

Hot deformation is one of the primary methods for fabricating anisotropic rare earth permanent magnets. Firstly, rapidly quenched powder flakes with a nanocrystal structure are condensed into fully dense isotropic precursors using the hot-pressing process. The prepared isotropic precursors are then hot-deformed to produce high-anisotropy uniaxial bulk rare earth permanent magnets and a highly textured structure is produced via this process. The resulting magnets possess many advantages such as near-net-shape, outstanding corrosion resistance, and ultrafine-grain structure. The influence of the preparation parameters utilized in the hot-pressing and deformation processes on the magnetic properties and microstructure of the permanent magnets are systemically summarized in this report. As a near-net-shape technique, the hot deformation process has notable advantages with regard to the production of irregular shapes, especially for radially oriented ring-shaped magnets with high length-diameter ratios or thin walls. The difficulties associated with the fabrication of crack-free, homogeneous, and non-decentered ring-shaped magnets are substantially resolved through an emphasis on mold design, adjustment of deformation parameters, and application of theoretical simulation. Considering the characteristics of hot-deformed magnets which include grain shape and size, anisotropic distribution of intergranular phases, etc., investigation and improvement of the mechanical and electric properties, in addition to thermal stability, with the objective of improving the application of hot-deformed magnets or ring-shaped magnets, is of practical significance.

1. Introduction

In the mid-1980s, Lee[1] originally reported on a method to prepare an anisotropic Nd–Fe–B magnet with ultrafine grain size by hot pressing and die-upsetting the rapidly quenched Nd–Fe–B alloy at a high temperature. Compared to the traditional sintering method, the die-upsetting process can realize a strong magnetic anisotropy without the application of an external magnetic field. It is therefore appropriate for the production of ring magnets with radial orientation. The formation of texture in this kind of Nd–Fe–B magnet, i.e., hot deformation (HD) magnet, is attributed to the anisotropy of the elastic modulus of the Nd2Fe14B grains. As shown in Fig. 1,[2] the elastic modulus along the c-axis (E ) is much smaller than that perpendicular to the c-axis (E ). In addition, the difference increases rapidly when the temperature is higher than 500 °C. Therefore, the c-axes of the Nd2Fe14B grains tend to be parallel to the pressure direction at high temperature. In fact, the detailed mechanism of alignment or texture formation is more complicated in a real system. It is generally acknowledged that the high anisotropy is attributed to both the rotation and the preferred growth of the Nd2Fe14B grains under pressure at high temperature. The grains, whose c-axes form smaller angles with the press direction, can more easily devour the misoriented grains via the precipitation growth or diffusion slip processes, because the strain energy of the misoriented grains is much higher than that of the well-aligned grains.[3,4] The preferred growth and rotation of the grains may be simultaneous, or more likely, they dominate in different steps in the texture formation process. As a special example, the radial alignment for the back extruded ring magnet is obtained via the platelet-shaped grain rotation or alignment in the extrusion process, and these grains are formed before extrusion.[5] Obviously, rare-earth-rich intergranular phase (liquid phase at high temperature), i.e., rare earth content greater than the stoichiometric ratio of Nd2Fe14B, is an essential factor in the process of obtaining platelet-like shape grains and the magnetic anisotropy of the whole magnet. Of course, the plastic deformation of the magnet in the die upsetting process is also necessary. Figure 2 shows the demagnetization curves of the over-stoichiometric Nd13.34Fe74.64Co5.5Ga0.42B6.0Nb0.1 and rare-earth-lean Nd7.91Pr2.68Fe84.01B5.39 (at.%) hot pressing (HP) and HD magnets.

Fig. 1. Variations of elastic modulus of Nd2Fe14B along (E ) and perpendicular (E ) to c-axis directions and their difference with temperature. The data is extracted from Ref. [2].
Fig. 2. Demagnetization curves of HP and HD for the over-stoichiometric and rare-earth-lean magnets.[6]

There are three advantages of the hot deformation process based on the conclusions of Croat et al.:[7] fewer manufacturing steps and therefore suitable for continuous mass production; net shaping or near net shaping which requires minor or no finish grinding; ideal process for producing specially shaped magnet such as arc or ring shape with radial orientation. To date, backward extruded ring magnets with radial orientation have been used in electric power steering systems and in AC servo motors. HD magnets with their ultrafine-grain structure have the potential to yield a high coercivity (H cj), and as a result, they have regained much attention. Currently, most of the investigations on hot deformed Nd–Fe–B magnets are primarily focused on the preparation technology and microstructure optimization to improve magnetic properties, especially coercivity. Moreover, the stress-induced crystallographic orientation is another critical issue pertaining to hot-deformed Nd–Fe–B magnets. Hot-pressing and hot deformation processes have prominent advantages with respect to the production of bulk nanocomposite permanent magnets, and were once considered to be candidates for the fourth generation of rare earth permanent magnets. However, the results of previous studies have indicated that texture formation is very difficult to achieve using a hot deformation process in the case of nanocomposite magnets due to the rare-earth-lean composition. By infiltrating the grain boundary of nanocomposite magnets with eutectic alloys, the capacity for deformation and texture formation are significantly improved. The orientation mechanism of diffusion-processed nanocomposite magnets has also been extensively discussed in recent years.

2. Effects of preparation parameters on HD magnets

The preparation process including melt spinning, hot-pressing, and hot deformation significantly influences the microstructure and magnetic properties of HD magnets. As the starting material, the powders crushed from melt-spun ribbons consist of nanocrystals[8] or amorphous matrix.[9] Although melt-spun ribbons exhibit high oxidation resistance, the oxygen content of the HD magnet increases with a decrease of the size of the particles in the powders. In order to produce hot deformed magnets with better performance, the particle size of commercial melt-spun powders is generally . Moreover, the microstructure and magnetic properties of hot-deformed magnets are closely related to the hot pressing and hot deformation process. Herein, the influence of the preparation parameters on the microstructure and magnetic properties of magnets will be discussed in detail, such as hot pressing temperature, hot deformation temperature, deformation degree, and strain rate.

2.1. HP temperature

An isotropic precursor is obtained by hot-pressing melt-spun powders. The density variation of the precursors prepared at different HP temperatures is shown in Fig. 3(a). The density monotonically increases with an increase of the HP temperature before 670 °C and reaches near full density (∼7.6 g/cm3) with a further increase in temperature. Figure 3(b) illustrates the demagnetization curves of HP magnets prepared at different temperatures. When the HP temperature increases from 630 °C to 710 °C, the remanence B r remains at approximately 8 kG because of the effect of magnetic isotropy. The squareness of the demagnetization curve of the sample prepared at 630 °C is distinctly inferior to that of the curves at 670 °C and 710 °C. This can be readily accounted for by the density variation. The coercivity of the samples prepared at 630 °C and 670 °C is ∼18.5 kOe, which is slightly higher than the value of the sample prepared at 710 °C. The microstructure of the HP precursor is sensitive to elevated temperatures while the size of the grains remains at 30–80 nm.[10] An isotropic fully dense precursor with small grains is favorable to the magnetic properties.[11] A further elevated HP temperature results in the formation of coarse grains and a consequent reduction of the hot workability, which seriously weaken the c-axis texture in HD magnets. Li et al.[12] reported that the microstructure of HD magnets is closely related to the initial microstructure of their precursors. In addition, Liu et al.[13] noted that magnets prepared at elevated temperature may give rise to the extrusion of partial rare-earth-rich phases, which reduces the coercivity because of the weakened decoupling between grains.

Fig. 3. (a) Dependence of the density of HP precursor on HP temperature ranging from 590 °C to 710 °C and (b) demagnetization curves of HP precursors prepared at the temperatures of 630 °C, 670 °C, and 710 °C.[6]
2.2. HD temperature

The fully dense precursors are deformed to be anisotropic magnets with good texture during the HD process under suitable temperature and stress.[10] The texture in HD Nd–Fe–B magnets develops because of the stress-induced preferential grain growth via a dissolution-precipitation mechanism.[14,15] Figure 4 shows the microstructure of hot-deformed magnets prepared at different HD temperatures from 780 °C to 860 °C, as observed by scanning electron microscope (SEM). For the magnet deformed at 780 °C, equiaxed grains and fine platelet-shaped grains with inferior alignment coexist, as shown in Fig. 4(a). This indicates that both the growth and orientation of the Nd2Fe14B grains are inhibited during the HD process due to the relatively low temperature. When the HD temperature increases to 820 °C and 840 °C, well-aligned platelet-shaped grains with a thickness of 50–70 nm and a length of 300–700 nm are formed (Figs. 4(b) and 4(c)). A further increase of the HD temperature to 860 °C results in an obvious increase of the grain size (Fig. 4(d)).

Fig. 4. SEM photos of HD magnets prepared at the HD temperature of (a) 780 °C, (b) 820 °C, (c) 840 °C, and (d) 860 °C. The white arrows represent the easy magnetization direction (c-axis) which is perpendicular to the plane of platelet grains.

Demagnetization curves of HD magnets prepared at the HD temperatures of 780–860 °C are shown in Fig. 5. It is seen that the remanence increases with the HD temperature before 840 °C and then decreases with a further increase of the HD temperature to 860 °C. For the magnet deformed at 780 °C, the low remanence is due to the stronger deformation resistance and consequent lower orientation. When the HD temperature is increased to 820–840 °C, the platelet-shaped Nd2Fe14B grains are well-aligned. This is consistent with the higher value of the remanence. The coercivity decreases monotonically with an increase of the HD temperature because of the grain growth. Based on the preceding analysis, the deformation temperature was optimized to be in the range 820–840 °C. The remanence, coercivity, and maximum energy product of the HD magnet deformed at 840 °C are 13.8 kG, 15.8 kOe, and 46.5 MGOe, respectively. Lin et al.[16] reported that abnormal grain growth and Nd-rich phase extrusion occur when the HD temperature is higher than 850 °C, which leads to a reduction of the density and remanence. Zhao et al.[17] reported that the effective anisotropy can be obtained at the HD temperature of 600–900 °C. In their work, it was shown that a suitable HD temperature was about 800 °C. It should be noted that the suitable HD temperature varies based on the different heating methods. The optimum HD temperature of 680 °C has been reported by Hu et al.[18] using spark plasma sintering as the deformation method.

Fig. 5. Demagnetization curves of HD magnets prepared at different HD temperatures.
2.3. Deformation degree

Hot deformed magnets with deformations ranging from 30% to 80% were produced and their magnetic properties were investigated (Fig. 6(a)). When the deformation increased from 30% to 70%, the coercivity decreased from 17.3 kOe to 15.3 kOe while the remanence increased from 10.6 kGs to 13.75 kGs. A further increase in the deformation up to 80% resulted in a rapid decrease of the coercivity and the initiation of a decline of the remanence. The maximum energy product presented a similar variation with the remanence, and the maximum value was at a deformation of 70%. X-ray diffraction (XRD) patterns of the HD magnets with deformations of 30%, 70%, and 80% are shown in Fig. 6(b). For the 30%-deformation magnet, the intensities of the (004), (006), and (008) peaks are relatively weak, indicating an inferior texture. When the deformation is in the range of 70%–80%, the intensity of the (001) peak increases significantly. The (006) peak shows a higher intensity than the (105) peak, which indicates a strong texture in the HD magnet. Larger remanence and maximum energy product are obtained in HD magnets with higher deformation because of the enhancement of the texture.

Fig. 6. (a) Magnetic properties of hot deformed magnets with deformation of 30%, 52%, 64%, 70%, and 80%. (b) XRD patterns of the hot deformed magnets with deformation of 30%, 70%, and 80%.[6]

Figure 7 shows the microstructure of hot deformed magnets with the deformation degree of 30%, 70%, and 80%. In the case of the 30%-deformation magnet, the equiaxed grains and the platelet-shaped grains with poor alignment coexist due to insufficient deformation (Fig. 7(a)). In the 70%-deformation magnet, fine platelet-shaped grains with superior alignment are formed, as shown in Fig. 7(b). When the deformation increases to 80%, the platelet-shaped grains still exhibit fine alignment but the grain size is up to in length (Fig. 7(c)), resulting in a significant reduction of the coercivity. Based on these experimental results, it is concluded that the best deformation for the optimal magnetic properties is approximately 70%. However, the optimized deformation was determined to be 75% in Zhao’s research[17] and 80% in Liu’s.[19] In Zhao’s work, a higher deformation leads a non-uniform outcome and defects such as microcracks between melt spun ribbons in the magnet.[17] When well-aligned Nd2Fe14B platelets are realized in an HD magnet, it is difficult to further enhance the remanence by merely increasing the degree of deformation. In addition, the optimal deformation degree of an HD magnet typically depends on the composition.

Fig. 7. Microstructure of hot deformed magnets with deformation of (a) 30%, (b) 70%, and (c) 80%. The white arrows represent the easy magnetization direction of the c-axis that is perpendicular to the plane of the platelet grains.[6]
2.4. Strain rate

The deformation speed can be evaluated with a constant strain rate , where , h 0 is the initial height of the HP precursor and h is the ultimate height of the hot deformed magnets.[17,20] The strain rate significantly affects the microstructure and magnetic properties of hot deformed magnets. Figure 8 represents the dependence of the magnetic properties of the hot-deformed magnets on the strain rate. The coercivity increases monotonically with the increment of the strain rate. This effect is ascribed to the influence of finer grains at higher deformation speed due to the shorter high temperature exposure time. Both the remanence and maximum energy product increase initially, followed by a rapid decrease with the increase of the strain rate. The best strain rate was shown to be 25 ×10−3 s−1. Higher strain rates lead to less deformation of the grains and inferior texture. Zhao et al.[17] reported that the maximum energy product and remanence were obtained at a strain rate of 10−2 s−1. This suggests that a lower strain rate causes a longer deformation time and results in a lower coercivity for grain coarsening. In fact, a reduction of the remanence occurs if the deformation is too slow because the non-aligned coarse grains prevent the formation of c-axis texture which originates from fine grains.

Fig. 8. (color online) Magnetic properties of hot deformed magnets prepared with different strain rates.
2.5. Inhomogeneity of microstructure in HD magnet

A coarse grain region and fine equiaxed grain region can be found in hot deformed magnets as shown in Figs. 9(a) and 9(b). Both of these regions grow rapidly with the increase of the deformation temperature[16] and degree of deformation.[21] The two kinds of grains are disadvantageous with respect to magnetic properties. In particular, the coarse grains can lower the remanence. The coarse grain regions contain higher Nd content and mainly form at the ribbon boundaries with quasi-periodic distribution. Li et al.[12] reported that the growth of coarse grains occurs in the HP precursor at high hot compaction temperature and propagates during the subsequent hot deformation. Lai et al.[22] reported that the coarse grain region is correlated to the incomplete contact interfaces of neighboring ribbons. Under high temperature and stress, the Nd-rich phase is squeezed into the interspaces of neighbor ribbons and stress concentration exists in the contact points. Non-aligned coarse grain regions form in both HP and HD processes because of the fluidity of the Nd-rich phases and the release of interfacial energy.

Fig. 9. SEM images of (a) coarse grain regions and (b) fine equiaxed grain regions.

Figure 10 shows further results on the quantitative variation of magnetic properties along the axial direction over the HD magnets with different deformation degrees. The remanence of samples 2 and 3 is significantly higher than that of samples 1 and 4. The inhomogeneity of the remanence is indicative of the inhomogeneity of deformation along the axial direction in the HD magnets. In addition, the inhomogeneity of the microstructure also exists along the radial direction of the magnet.[23]

Fig. 10. (color online) Distribution of remanence (B r), coercivity (H cj) and maximum energy product ((BH)max) along axial direction of the HD magnet with deformation of (a) 37% and (b) 52%.[23]
2.6. Influence of chemical composition

Hot deformed Nd–Fe–B magnets are made of multi-phase materials that consist of the Nd2Fe14B matrix phase, Nd-rich phase, and B-rich compound. Magnetic properties, especially remanence and energy product, strongly depend on the grains alignment along the c-axis direction and the content and distribution of the Nd-rich phase. In the deformation process, Nd2Fe14B grains are wetted by the liquid Nd-rich phase, which promotes the formation of platelet-shaped grains via grain slipping and rotation.[3] In addition, Nd-rich phase at grain boundaries enhances the decoupling effect of Nd2Fe14B grains and facilitates the augmentation of the coercivity. Therefore, the Nd content in a hot deformed magnet is higher than that in the Nd2Fe14B tetragonal phase. Yi et al.[4] investigated the influence of the Nd content on the microstructure and magnetic properties of hot deformed magnets with nominal composition Nd12.8+xFe75.28−xCo5.5Ga0.42B6, as shown in Figs. 11 and 12. As the Nd content was increased, the remanence and energy product also increased until , followed by a reduction (Fig. 11). The coercivity increased monotonically with the addition of x from 0 to 1. As shown in Fig. 12(a), it is evident that equiaxed grains exist in the Nd-lean alloy (x = 0). Figure 12(b) shows that the main grains are elongated, and a good texture is obtained in the Nd-rich magnet for x = 0.54. However, the parallel alignment of the Nd2Fe14B grains is strongly distorted and circular alignment occurs in the excessive Nd-rich alloy at x = 1.08, as shown in Fig. 12(c). Similarly, Liu[13] produced Nd–Fe–B magnets with a coercivity of 17.9 kOe by increasing the Nd content and revealed that the pinning mechanism plays a dominant role in determining the coercivity. In Nd–Fe–B magnetic materials, Dy[24] and Tb[25] are the most commonly used additives to improve the coercivity due to their superior magnetocrystalline anisotropy fields (H a). Tang et al.[26] increased the coercivity of the Nd–Fe–B hot-deformed magnet from 15.0 kOe to 22.7 kOe by the addition of Dy-based melt-spun powders.

Fig. 11. Magnetic properties of the Nd12.8+xFe75.28−xCo5.5Ga0.42B6 magnets with different Nd content.[4]
Fig. 12. (color online) Microstructure of Nd12.8+xFe75.28−xCo5.5Ga0.42B6 magnets with (a) x = 0, (b) x=0.54, and (c) x=0.18.[4]

It has been reported that coercivity and maximum energy product are enhanced significantly by the addition of Cu powder using dual-alloy technology.[14] Microstructure analyses indicate that the NdCu2 phase is formed in grain boundaries, which obviously improves the texture of the Cu-doped magnet due to the lower melting point of the intergranular phase. This leads to the enhancement of magnetic properties. Ma’s research[27] indicates that Nb addition leads to a notable increase in both the remanence and the coercivity because of the improvement of the c-axis texture and the refinement of the microstructure. Microstructure analysis revealed that Nb atoms are enriched at the grain boundaries and form NbFeB compounds. Li et al.[28] researched the influence of Al, MoS2, Zn, etc. on magnetic properties, and found that MoS2 and Al are effective in increasing the remanence and energy product, while Zn can increase the coercivity. Moreover, Co, Ga, Zr, V, Mo, etc. are utilized for HD magnets to tune the microstructure or improve the magnetic properties.[2931]

In fact, SmCo5 permanent magnets can also be fabricated by hot pressing and subsequent hot deformation. In recent years, Yue[32,33] reported on hot-deformed anisotropic SmCo5 magnets with a high coercivity over 5 T. The raw powder was prepared using a ball milling technique and the result revealed that deformation levels up to 90% lead to the formation of well-aligned platelet grains. Images of the microstructure are shown in Fig. 13.[32] As observed in Fig. 13(a), the size of the equiaxial grain is approximately 20 nm in the HP sample. After deformation, the grains grow and become elongated perpendicular to the press direction (arrows in Figs. 13(b) and 13(c)). However, the partial equiaxial grains still exist in the 70%-deformation magnet due to inadequate deformation. When the deformation increases to 90%, the hot deformed SmCo5 magnets mainly consists of aligned platelets of SmCo5 grains but the coercivity decreases sharply as low as 0.58 T. Therefore, deformation plays a key role in inducing grain alignment and the formation of platelet shape.

Fig. 13. TEM images of (a) hot pressed, (b) 70%-deformation, and (c) 90%-deformation SmCo5 magnets.[34]
3. Coercivity enhancement for HD Nd–Fe–B magnets

Coercivity is an intrinsic property which plays a key role in the commercial application of Nd–Fe–B magnets. However, the maximum experimental value of coercivity is only 20%–30% of the theoretical value deduced according to the S-W model.[34] Enhancement of coercivity has been a primary challenge for researchers who study Nd–Fe–B magnets. The coercivity of Nd–Fe–B magnets is sensitive to the microstructure including defects on the surface of grains, ferromagnetic grain boundary phases, and effective demagnetization factors. In order to identify an appropriate method for improving the coercivity, there has been a significant effort directed at attempting to understand the coercivity mechanism of Nd–Fe–B magnets.

3.1. Coercivity mechanism of HD Nd–Fe–B magnets

Generally, there are two dominant coercivity mechanisms of Nd–Fe–B magnets: nucleation and domain wall pinning mechanisms. However, a consensus has not been achieved with regard to the dispute of the validity of the two mechanisms. In hot-deformed Nd–Fe–B magnets, a strong domain wall pinning is considered to be the primary coercivity mechanism.[35,36] Theoretically, the strength of the domain wall pinning force at the grain boundaries determines the coercivity values. The initial magnetization curves are often adopted to evaluate the magnetization behavior and coercivity mechanism. An S-shaped profile of the initial magnetization is obtained for the HD magnets, which indicates the existence of complicity in the coercivity mechanism.

The first step shows a high reversal susceptibility and relative ease of magnetization. It indicates that reversal domain wall motion dominates the magnetization behavior. In the second step, domain wall pinning at the grain boundaries is dominant. The two-step S-shaped magnetization curve indicates an inhomogeneous domain wall pinning mechanism in the HD Nd–Fe–B magnets. In addition, Volkov et al.[36] investigated the magnetic domain walls structures in hot-deformed Nd–Fe–B magnets and discussed domain wall pinning at the grain boundaries and inhomogeneous microstructures in detail. The inhomogeneous microstructures, which are grain boundary phases and different types of defects on the grain surfaces, result in the inhomogeneous domain wall pinning. An inhomogeneous domain wall pinning mechanism can be used to describe the dominant coercivity mechanism in HD Nd–Fe–B magnets. However, a nucleated micromagnetic model ( ) is always applied to analyze the coercivity mechanism in HD Nd–Fe–B magnets. A linear relationship between and is obtained so that the coercivity mechanism of hot-deformed magnets appears to be similar to the nucleation mechanism as well.

The enhancement of the coercivity has been the primary challenge for researchers. In Nd–Fe–B magnets, it has been reported that the coercivity increases with a decrease of the grain size. According to the S-W model, the maximum coercivity could be obtained as soon as the grain size is single domain size, and the coherent rotation of the magnetic moment is perfectly achieved. Due to their nanocrystalline grains, HD Nd–Fe–B magnets have a potential to achieve a high coercivity. However, the coercivity of these magnets is significantly lower than the theoretical value, even though the grain sizes are so-called nanoscale. This may be caused by ferromagnetic intergranular phases, platelet-shaped grains, surface defects, etc., which limit the enhancement of the coercivity for hot-deformed Nd–Fe–B magnets. Reversal domains would be nucleated at ferromagnetic intergranular phases and grain surface defects, followed by propagation into Nd2Fe14B grains. Moreover, the platelet-shaped grains may also be the primary cause of the relatively low coercivity which would significantly increase the effective demagnetization factor N eff.

3.2. Coercivity enhancement of hot-deformed Nd–Fe–B magnets

Although many of the aforementioned disadvantages could not be avoided in HD Nd–Fe–B magnets, significant effort has been made to increase the coercivity. The coercivity enhancement mechanism is discussed in this section.

The addition of certain elements is beneficial to the enhancement of coercivity in HD Nd–Fe–B magnets.[30,37,38] It is commonly assumed that this enhancement is attributable to the modification of the grain-boundary phase by the addition of the elements. Moreover, microstructure optimization and enhancement of magnetocrystalline anisotropy are both beneficial to the increase of the coercivity. Rare earth fluorides have been utilized to produce 45 EH and 50 UH grade sintered Nd–Fe–B magnets for practical application based on their resistance to oxidation and the operational convenient of the procedure. In HD Nd–Fe–B magnets, rare earth fluorides have also been utilized in the fabrication of high coercivity magnets. Tang et al.[39] and Sawatzki et al.[40] have independently reported on the use of DyF3 compounds to produce hot compact precursors followed by hot-deformation. The substitution of Dy for a portion of Nd results in the formation of (Nd, Dy)2Fe14B, and is considered to be the main reason for the enhancement of the coercivity. However, the existence of a mass of DyFe3 compounds in the diffusion regions between adjacent ribbons leads to a large wastage of rare earth resources and a decrease in remanence. Moreover, the scarcity of heavy rare earth (HRE) elements has motivated many researchers to investigate a more effective way to increase coercivity without or with minimal use of HRE. Microstructure optimization may prove to be a more effective way to increase coercivity. In a recent study, Zheng et al.[41] discovered that the coercivity can be increased from 15.6 kOe to 17.75 kOe by introducing WC alloy with a high melting point to inhibit the growth of coarse grains. Figure 14 shows SEM-BSE images of HD samples without and with WC addition. It can be seen from Fig. 14(d) that the coarse grains are refined by the addition of WC compounds compared to the initial hot-deformed Nd–Fe–B magnets, as shown in Fig. 14(c). The inhibition of coarse grain growth is a potential avenue to the development of high coercivity Nd–Fe–B magnets.

Fig. 14. (color online) (a) BSE images and (b) the EDS elemental map of W obtained from the 1.0 WC sample. (c) and (d) BSE images of coarse grains region for the initial and 1.0 WC sample, respectively.[41]

Grain refinement and the optimization of grain boundaries can eliminate ferromagnetic phases and surface defects, and are among the most effective procedures for increasing the coercivity. It is well-accepted that the coercivity of Nd–Fe–B magnets is influenced by the content of Nd-rich phases, and the coercivity increases with an increase in the content of Nd-rich phases.[13] Recently, grain boundary diffusion of rare earth eutectic alloys in nanostructure magnets has been investigated to increase coercivity. It has been shown that HDDR powders with a coercivity of 19.5 kOe can be obtained using a diffusion process based on Nd–Cu eutectic alloy.[42] The thickened grain boundary of the HDDR powders is considered to be the primary reason for the enhanced coercivity. The coercivity of melt-spun Nd–Fe–B powders was shown to be enhanced to 2.6 T due to Nd–Cu infiltration.[43] Sepehri-Amin et al. reported that the high coercivity of ultrafine-grained hot-deformed Nd–Fe–B magnets with 2.3 T is due to the formation of Nd–Cu GBDP.[44] An analysis of the element distribution proved that the ferromagnetic elements in the intergranular phases are reduced from ∼55 at.% to an undetectable level, which results in the decoupling of Nd–Fe–B grains and an increase of the coercivity.

In addition, a much higher coercivity at 2.6 T is simultaneously achieved by low-temperature grain boundary diffusion of Nd60Dy20Cu20 eutectic alloy and Pr70Cu30 alloy, respectively.[4547] Many researchers have determined that the composition of ferromagnetic elements in intergranular phases parallel to the c-axis is substantially reduced and that a thin Dy-rich shell surrounds Nd2Fe14B grains. Based on previous investigations, an optimized composition of PrNd–Cu alloy was utilized and an HD Nd–Fe–B magnet with a higher coercivity of 2.7 T was obtained.[48] The BSE SEM images of the HD and diffusion-processed magnets are illustrated in Fig. 15. The images of the diffusion-processed magnets with the c-axis in-plane and out-of-plane are shown in Figs. 15(b) and 15(c), respectively. The Nd-rich intergranular phases became thicker after the eutectic alloy diffusion treatment.

Fig. 15. TEM-EDS images of the (a) hot-deformed and (b), (c) diffusion-processed magnets.

Based on the demagnetization profiles presented in Ref. [48], the coercivity of the surface regions is stronger than that of the center regions of the diffusion-processed magnets. Moreover, the TEM images in Ref. [48] indicate that the diffusion process is inhomogeneous, which results in magnetization behavior similar to nanocomposite magnets. Recently, one of the highest reported coercivity values of 28 kOe was achieved without HRE addition by diffusion of low melting temperature glass forming Pr60Al10Ni10Cu20 alloys.[49] The large enhancement of the coercivity is attributed to the magnetic isolation of the Nd2Fe14B grains by the Nd-rich amorphous/crystalline intergranular phase. These phases are considered to be non-magnetic. Bance et al.[50] reported that the thickness of surface defects, which are commonly considered to possess lower magnetocrystalline anisotropy, plays a critical role in the magnetization reversal process. In addition, it was found in Ref. [50] that the single-grain computed results for the angular dependence of the coercive field appear to be in good agreement with experimental results when imperfect grain alignment of real magnets is considered. Nonetheless, the enhancement of the coercivity is primarily due to the modification of grain boundaries by the infiltration process of rare earth eutectic alloys.

Figure 16 represents a summary of the currently developing situation with respect to the coercivity in HD Nd–Fe–B magnets according to selected references. It can be seen that the coercivity can be significantly enhanced via the application of an appropriate optimization diffusion process. However, simultaneously, the remanence is dramatically decreased. Generally, the remanence (M r) could be described as follows:

where A is the volume fraction of the positive domain parallel to the pre-magnetization, ( ) is the volume fraction of the hard magnetic phase, d 0 is the theoretical density, d is the actual density, is the orientation degree, and M s is the saturation magnetization of the main magnetic phase. After the diffusion treatment, the excessive non-magnetic phases will increase the value of β, resulting in a decrease of the remanence. In fact, the disruption of the grain orientation, which may be a neglected key factor, results in a decrease of the remanence and energy production. In order to achieve a high coercivity without sacrificing the remanence, Akiya et al.[51] introduced an effective method based on low eutectic Nd–Cu diffusion under expansion constraint. However, the enhancement of the coercivity was limited.

Fig. 16. (color online) Current development of coercivity in HD Nd–Fe–B magnets.

In sintered Nd–Fe–B magnets, optimized structures of (Nd, Dy)2Fe14B or (Nd, Tb)2Fe14B surrounding the Nd2Fe14B phases can be formed using GBDP to yield a dramatic increase in the coercivity without remarkable reduction in the remanence. In hot-deformed Nd–Fe–B magnets, a high coercivity may also be obtained if such a core-shell structure is achieved. Liu et al.[54] reported on a diffusion-processed HD magnet with a coercivity value of 2.75 T and a remanence value of 1.3 T using Nd62Dy20Al18 alloys as a diffusate. In this study, both the core-shell structure with a high magnetocrystalline anisotropy, and magnetic isolation decoupled by nonmagnetic intergranular phases, result in a large enhancement of the coercivity with a slight decrease of the remanence. Recently, Wang et al.[53] developed a procedure to obtain excellent comprehensive magnetic properties of HD Nd–Fe–B magnets based on Dy-Cu press injection and subsequent annealing. The demagnetization curves and magnetic properties of the hot-deformed and Dy–Cu press injected magnets at different stages of the process are shown in Fig. 17. High performance can be achieved by optimized annealing while the near-surface microstructure is remarkably improved and the core-shell structure is formed by Dy–Cu press injection.

Fig. 17. (color online) (a) Demagnetization curves of the hot-deformed and Dy–Cu press injected magnets. (b) Magnetic properties of the hot-deformed and Dy–Cu press injected magnets in different stages of the procedures: initial state, S-step, and D-step annealed states.[53]
3.3. Discussion on the coercivity enhancement of HD Nd–Fe–B magnets

Based on the preceding discussion, the coercivity is closely related to facet defects of grains and the magnetism of the intergranular phases. A high coercivity can be obtained by modification of the grain boundaries and the intergranular phases. Sepehri-Amin et al.[44] investigated the influence of the magnetism of intergranular phases on the coercivity using a finite element model. The micromagnetic simulation result indicates that a reduction of the magnetization in the intergranular phase would effectively improve the coercivity, which is in agreement with experimental results.

Moreover, according to the S-W model, magnetic isolation by the nonferromagnetic intergranular phases is also beneficial for the enhancement of coercivity in HD Nd–Fe–B magnets. Bance et al.[50] studied the influence of defect thickness on the angle of the coercivity. The results indicate that the magnetization reversal mechanism shifts from nucleation to de-pinning with increasing defect thickness. Tang et al.[48] analyzed the magnetization behavior of the diffusion processed HD magnets and identified an S-shaped magnetization profile as shown in Fig. 18, which suggests a strong pinning behavior. Based on Fig. 19, the magnetization exhibits a 3-step process. Domain wall reversible motion within the Nd2Fe14B grains is dominant in the first step, while domain wall pinning and magnetization reversal of the grains are the primary magnetization processes of the second step. As a result, the coercivity may be dramatically enhanced if the grains are perfectly magnetically isolated by non-ferromagnetic phases, and the coherent rotation can be achieved. However, Tang et al. also reported that the distribution of low eutectic alloys is actually inhomogeneous, and some grains are surrounded by infiltrated phases, while other grains are still stacked together. This may be because the distribution of grain boundaries is inhomogeneous in initial HD Nd–Fe–B magnets. The inhomogeneous distribution of Nd-rich phases results in the open recoil loops in Fig. 15(a), which is not beneficial for practical use in diffusion processed magnets.

Fig. 18. (color online) Micromagnetic simulations of intergranular phases by finite element micro-magnetic simulation model (supplied from NIMTE).
Fig. 19. (a) Recoil loops of the diffusion processed sample; (b) field dependence of remanence and reversible susceptibility.[48]

In summary, the coercivity has become the dominant obstacle to the widespread application of Nd–Fe–B magnets. Researchers are faced with the difficult task of achieving a high coercivity (HRE-free) without reducing the remanence. Based on the results of previous investigations, a high coercivity may be obtained by appropriate optimizations of the grain boundaries and intergranular phases. Firstly, the Nd-rich intergranular phases, which are simply considered as nonmagnetic phases, should be the key microstructure for the coercivity enhancement of Nd–Fe–B magnets. If the ferromagnetism of the intergranular phases could be weakened and the neighboring grains could be decoupled, a high coercivity would be achievable. Secondly, the core-shell structure, which could inhibit the reversal domain nucleation due to the magnetic hardness of the surfaces of the grain, may also provide an effective avenue for obtaining a high coercivity. A high coercivity without a corresponding reduction of the remanence has been achieved in thin films which have an almost ideal microstructure.[54,55] Nonetheless, this is a great challenge and it is expected that a higher coercivity than that of thin films is required.

4. HD nanocomposite Nd–Fe–B magnets

Recently, new rare-earth permanent magnetic materials with high magnetic properties have attracted much interest due to the increasing demand for environment-friendly and energy-saving devices. However, the scarcity and high cost of rare earth (RE) limit the applications of high-performance rare-earth compounds, such as in the case of Nd–Fe–B permanent magnets. The relatively high concentration of iron (cobalt) atom of nanocomposite magnets has resulted in a reduction of the usage of RE elements, and thus makes a great contribution to the sustainable development of the RE industry. Therefore, the development of nanocomposite magnets has become one of the main avenues for easing the RE shortage. In addition to the reduced RE content, a high soft-magnetic phase fraction in nanocomposite magnets contributes to a large saturation magnetization for potentially high energy products. This implies that the exchange interaction between “soft-hard” magnetic phases can perform well if the “soft-magnetic” phase is thinner than the domain wall thickness of only 3.9 nm for the Nd2Fe14B system. The magnetostatic interaction is demonstrated to relax the strict requirements for the size of the soft-magnetic inclusions. To date, numerous soft-hard composite systems have been investigated. Currently, bonded Nd–Fe–B nanocomposite magnets fabricated by Nd–Fe–B/α-Fe magnetic powders have been widely used. However, the isotropic magnetic characterization and high concentration of nonmagnetic resin adhesive result in a substantial deterioration of the magnetic performance. The obtained maximum energy product (BH)max of the isotropic powders was lower than . Hence, the determination of a procedure for the fabrication of high-anisotropy nanocomposite magnets has become a significant issue for both industry and academia.[56]

The hot-pressed/hot-deformed technique has become an important branching procedure for preparing fully dense anisotropic nanocrystalline magnets. Generally, we can obtain anisotropy hot-deformed Nd–Fe–B composite magnets by several different pathways. The nanocomposite Nd–Fe–B magnets were prepared with isotropic Nd-lean magnetic powders. Compared to the Nd-rich magnetic powders, the Nd-lean magnetic powders give rise to the degradation of the deformability for the bulk material. It is difficult to realize a stress-induced crystallographic orientation in single nanocomposite Nd–Fe–B magnets. Subsequently, the co-doping technique has been widely used to improve the deformability of nanocomposite magnets, such as Nd-rich/Nd-lean[57,58] and Nd-rich/Fe(Co) systems.[59] Since low-melting RE-rich intergranular phase plays an essential role in the stress-induced solution-precipitation creep process,[15,60,61] RE-rich eutectic can be diffused into the grain boundaries to facilitate the formation of a highly textured structure.

4.1. Nanocomposite magnets prepared by Nd-lean powders

To obtain the Nd-lean magnetic powders with rare-earth content lower than the chemical stoichiometry, the ingots with lower RE contents are usually quenched into melt-spun ribbons.[6265] Then the nanocomposite magnets fabricated by these crystallized Nd-lean magnetic powders are obtained using the hot-pressed and hot-deformed processes. The demagnetization curves of nanocomposite Nd–Fe–B hot-deformed magnets (nominal composition Nd Pr1Dy0.3Fe77.3Al0.2Co6.1Ga0.2B5.7 at.%) are measured parallel and perpendicular to the easy-magnetization direction (Fig. 20). It can be determined that the nanocomposite magnets exhibit a weaker anisotropy because of the low [001]-oriented structure. The microstructure characterization indicates that the short-axes of the nanocrystals are preliminarily aligned along the loading direction (Fig. 21). However, a large number of nanocrystals show equiaxed morphology. By increasing the deformation temperature (i.e., 900 °C), more Nd2Fe14B nanocrystals could be shortened along the easy-axis and elongated along its perpendicular direction.[66] Considering the lack of Nd-rich phases, the textured structure is probably derived from shear deformation under the combined action of pressure and temperature. Specifically, the anisotropic elasticity modulus of Nd2Fe14B crystal accelerates the rotation and sliding of the lattice plane. It should be noted that the magnetic properties of the nanocomposite magnets decrease with an increase in the content of soft-magnetic phases. To date, the maximum energy product obtained in the laboratory is only in the range of . Apparently, the obtained unsatisfactory remanence and maximum energy product are due to the disordered crystallographic orientation. Improving the grain alignment in nanocomposite magnets has become a challenge in the development of high-performance RE permanent magnets. To address this challenge, the uniaxial ultrahigh stress is employed to induce the orientation of Nd-lean amorphous magnetic powders. This method has successfully yielded highly textured structures.[63,64,67]

Fig. 20. Demagnetization curves of rare-earth-lean Nd–Fe–B nanocomposite magnets parallel and perpendicular to easy-magnetization direction.[62]
Fig. 21. (a) SEM fracture surface of hot-deformed magnets with a composition of Nd9.3Pr1Dy0.3Fe77.4Co6.1Ga0.2B5.7 (at.%). (b) TEM micrograph and SAED pattern (inset) of magnets with a composition of Nd9.3Pr1Dy0.3Fe77.4Co6.1Ga0.2B5.7 (at.%).[62]

In amorphous systems, the crystallization control in amorphous alloys is of particular importance for the development of high-performance nanocrystalline materials. Li et al.[74] succeeded in controlling α-Fe and Nd2Fe14B crystallization processes in amorphous Nd9Fe85B6 by a combination of severe plastic deformation at room temperature and subsequent heat treatment, as shown in Fig. 22. The Nd2Fe14B/α-Fe nanocomposite magnets prepared by this approach possess homogeneous microstructures with a small size, 15 nm for α-Fe phase and 26 nm for Nd2Fe14B phase. Therefore, they show enhanced magnetic properties compared to those prepared by direct annealing of amorphous Nd9Fe85B6. The small initial phase size is prepared for the strong “hard–soft” coupling in the ultimate bulk magnets. To obtain a strong [001]-oriented texture, Nd-lean amorphous powders are hot-pressed and hot-deformed at a large uniaxial stress of 310 MPa. The unusual texture formation is attributed to a preferential nucleation of Nd2Fe14B crystals in an amorphous matrix induced by ultra-high stress. Figure 23 shows the platelet-shaped microstructure, in which Nd2Fe14B nanocrystals with sizes of 35–50 nm in diameter and 18–25 nm in thickness can be observed (Fig. 23). Selected area electron diffraction (SAED) reveals that the polycrystalline diffraction signals are due to the weak [001] texture caused by α-Fe phases. Due to the hash ultrahigh stress, other simple techniques are urgently required for the development of high-anisotropy nanocomposite magnets.

Fig. 22. TEM images and electron diffraction SAED patterns of amorphous Nd9Fe85B6 alloy (a) subjected to HPTD and subsequently annealed at 600 °C for 10 min and (b) directly annealed at 725 °C for 10 min.[67]
Fig. 23. TEM bright-field images (a) perpendicular and (b) parallel to the stress direction, (c) SAED patterns of the bulk amorphous Nd9Fe85B after hot deformation.[64]
4.2. Nanocomposite magnets prepared by the co-doping technique

Since Nd-lean magnetic powders exhibit poor deformability, co-doping techniques have been developed to fabricate high-performance nanocomposite magnets. For instance, Nd-rich magnetic powders are co-doped with Nd-lean magnetic powders to improve the deformability of bulk samples. Experimental results have demonstrated that the maximum energy product of co-doped magnets can be increased to .[59] This indicates that the Nd elements diffuse from the Nd-rich magnetic powders to the Nd-lean magnetic powders to improve the crystallographic orientation of the Nd-lean structure. Therefore, this method can realize the [001]-oriented texture in nanocomposite magnets. Moreover, the soft magnetic particles such as Fe–Co phase (Fig. 24(a)) are also blended with the Nd-rich powders, then subsequently hot deformed into bulk nanocomposite magnets. The BSE SEM image in Fig. 24(b) shows that the dark gray α-Fe phases are embedded in the brighter Nd–Fe–B matrix. In contrast to the nanocomposite structure, the α-Fe phase with a size of is far beyond the limitation of the short-range exchange coupling interaction. Apparently, the long-range magnetic interaction should play a critical role in the improved magnetic properties.[59] To further disperse the soft magnetic phases, they can be deposited on the surfaces of the Nd-rich powder. This method effectively increases the maximum energy product of hot-deformed magnets, up to the maximum value of .[68]

Fig. 24. (a) SEM micrograph of Fe–Co powder particles; (b) SEM back scattered electron image of a hot-deformed composite Nd13.5Fe80Ga0.5B6/α-Fe (91.7 wt.%/8.3 wt.%) magnet.[59]
4.3. The deformability enhancement of nanocomposite magnets by introducing rare-earth-rich liquid phase

Since the formation of texture in nanocomposite magnets is difficult, it is of importance to reveal the mechanism of texture formation in hot deformation. Prior investigations indicate that the RE-rich phase is an essential factor for the magnetʼs deformability. To investigate the deformation mechanism of rare-earth-lean nanocomposite magnets, the low-melting RE-rich eutectic alloy is added to the hot deformation process. Thus the formation of a textured structure can be controlled by adjusting the composition and concentration.

Figure 25(a) shows the microstructure of an isotropic precursor hot-pressed at 600 °C, where two adjacent powders and a narrow boundary area between two powder particles are observed. In terms of image contrast, there are at least three regions that can be acquired: (i) bright region; (ii) grey region; and (iii) dark region. The energy-dispersive x-ray spectrometry (EDS) analysis indicates that the bright region is the Nd–Cu-rich phase and the dark region corresponds to the raw nanocomposite particle without Nd–Cu diffusion; the presence of the transition grey region suggests that Nd–Cu diffusion takes place in this area. This result indicates that the Nd–Cu eutectic diffusion process is initiated during the hot-pressing process.

Fig. 25. (color online) (a) BSE SEM image showing the interface of particles in the sample hot pressed at 600 °C, (b) EDS maps of elements in this area and (c) EDS line-scan across the interface region marked in (a).[69]

Apart from the hot-pressed process, the mechanism of microstructure evolution in eutectic-doped hot-deformed magnets was also studied. The effect of low-melting Nd-rich intergranular phases on the formation of platelet-shaped grain and textured structure is critical. Tang et al.[70] studied the effect of Nd-rich phases on texture formation in nanocomposite magnets by the diffusion of Nd–Cu eutectic. The results demonstrate that the Nd-rich intergranular phases are favorable to the anisotropy of Nd–Fe–B magnets, as shown in Fig. 26. With the addition of appropriate Nd–Cu content, a well-aligned platelet-shaped structure can be obtained.

Fig. 26. Field emission SEM micrographs of hot-deformed nanocomposite magnets with various Nd–Cu alloy addition: (a) x = 0, (a) x = 4 wt.%, (c) x = 6 wt.%, and (d) x = 10 wt.%.[70]

In addition, the crystal structures in the intergranular phases are correlated with the chemical composition (Fig. 27). Tang et al.[69] studied the microstructure of intergranular phases in nanocomposite magnets, as represented by the results shown in Fig. 27. The HRTEM image of the grain boundary located at position 1, which is identified in Fig. 27(a), indicates evidence of intergranular phase. The fast Fourier transform (FFT) patterns of the identified regions show that the intergranular phases consist of a dhcp-Nd-rich phase and a hexagonal close-packed (hcp) Nd2O3 phase with lattice parameters of a = 3.83 Å and c = 5.99 Å. The decomposition from the face-centered cubic NdOx (fcc-NdOx) phase to the hcp-Nd2O3 and dhcp-Nd-rich phases occurs at temperatures below 863 °C. In this work, the relatively low hot deformation temperature of 850 °C creates favorable conditions for this decomposition. Furthermore, the dhcp-Nd-rich phase transforms easily into hcp-Nd2O3 via oxidation, because the latter is the most stable phase and the formation of hcp-Nd2O3 is energetically favored. After the addition of the Nd–Cu alloy, the coercivity of the hot-pressed magnets is significantly improved, which implies that the Nd–Cu liquid starts to diffuse into the powder during the hot-pressing process. Therefore, the subsequent high-temperature deformation is beneficial to the diffusion and texture formation process during hot deformation. The aforementioned findings reveal that the RE-rich intergranular phase plays a vital role in facilitating the formation of highly textured structures. The lack of RE-rich phases is the main reason for the poor deformability of nanocomposite magnets.

Fig. 27. TEM/HRTEM and fast Fourier transform images of hot-deformed nanocomposite magnets with a deformation rate of 30%.[69]
5. Radially oriented Nd–Fe–B ring magnet prepared by backward extrusion method

Radially oriented Nd–Fe–B ring magnet (RM) is a special Nd–Fe–B magnet with a ring shape with the easy c-axis along the radial direction (Fig. 28). Given that the RM is convenient for magnetization and assembly, it has great potential applications in many types of motors because of its special shape. So far, three kinds of Nd–Fe–B RMs have been produced: bonded RM, sintered RM, and hot-deformed RM. Bonded RMs are isotropic and possess a lower energy density than the two other RMs. Sintered RMs are fabricated via a complex procedure that is very time intensive and material-wasting. In contrast, hot-deformed RMs can be prepared via a facile near-net-shape method, which exhibits a significantly higher efficiency. Therefore, increasing attention has been focused on hot-deformed RM. As a unique hot deformation technique, backward extrusion is commonly used for the preparation of radially oriented NdFeB ring magnets. Moreover, backward-extruded RM has a high power-to-volume ratio, good corrosion resistance, and lower rare earth content. Therefore, it has great potential applications in servo motors, EPS motors, and driving motors.

Fig. 28. (color online) Schematic illustration of radially oriented Nd–Fe–B ring magnet.

The procedure of the backward extrusion method is illustrated in Fig. 29. It consists of two steps: hot-pressing and hot deformation (back extrusion). The HP step is employed to densify the melt-spun powders at a temperature in the range 600–700 °C under pressure. Since the melted Nd-rich phase exists in the Nd–Fe–B magnet at the HP temperature, it is easy to obtain a magnet with a nearly full density. HP magnet mainly consists of equiaxed Nd2Fe14B grains and thus exhibits isotropic magnetic performance. The HD step aims to produce ring shaping and texture formation at HD temperature (700–850 °C) in the extrusion process.

Fig. 29. (color online) Schematic illustration of the procedure of backward extrusion method.[6]
5.1. Texture evolution in backward extrusion of NdFeB ring magnet

The HD RM has significant potential for application in brushless DC motors.[71] However, some researchers have indicated that the inhomogeneity of magnetic performance limits the application of HD RM in industry to some extent. For example, the remanence decreased quasilinearly from the inner to outer surface of backwardextruded (BE) RM along the radial direction.[20,72] In the axial direction, remanence initially increased, then decreased from the top to the bottom of the RM.[73,74] To examine the cause of this inhomogeneity, the mechanisms of texturing and material flow must be evaluated.

To date, three main mechanisms have been proposed to explain the formation of platelet Nd2Fe14B grains in the hotdeformed magnet. Li et al.[60] suggested a mechanism based on the dissolution of nonfavorably oriented crystallites in the Nd-rich phase and their reprecipitation on the crystallites, which have their c-axes oriented parallel to the stress direction. An extensive study by Grünberger et al.[20] suggested that the diffusion creep mechanism played a dominant role, which is accompanied by rheological and viscous flow. Yuri and Leonowicz[75,76] have proposed another mechanism, which considered the mechanical rotation of grains when they deformed plastically, and the alignment results from the geometric anisotropy of the crystallites. However, the subsequent evolution of these Nd2Fe14B platelets has rarely been reported because they depend greatly on the preparation strategy.

Yin et al.[5] have performed a systematic investigation of texture evolution and materials flow over a partly deformed ring magnet. This ring magnet was divided into two parts: a ring-shaped part and a cylindrical part. Their investigation revealed that the formation of the platelet-shaped grains occurred in the cylindrical part and the depth of the texture zone was limited by the bottom of the punch. Further investigations suggest that equiaxed grains are distributed in the bottom region of the cylindrical part because no material flow occurs in that region (Fig. 30). Moreover, the orientation of the Nd2Fe14B platelets changes in the region where the ring-shaped part and the cylindrical part meet (Fig. 31), resulting in a 90° rotation of the c axis when the Nd2Fe14B platelets flow from the cylindrical part into the ring-shaped part.

Fig. 30. (color online) Schematic illustration of the texturing zone and the non-texture zone in the cylindrical part of the ring magnet prepared using the backward extrusion method.[5]
Fig. 31. SEM images of the region where the cylindrical part meets the ring-shaped part: (a) the image under low-magnification, (b)–(d) the high-magnification images of regions B, C, and D in (a), respectively. The white arrows represent the pressing direction.[5]
5.2. Magnetic performance improvement of backward-extruded NdFeB RM

Radially oriented RM exhibits inhomogeneity of the remanence along the axial direction.[72,77] This inhomogeneity has been ascribed to the texture difference in the axial direction.[78,79] To address this problem, Tang et al.[80] developed a pre-deformation technique. The HP magnet was initially subjected to a 52% deformation. This was followed by backward extrusion into the RM. The RM prepared using this technique exhibited a significant improvement in the homogeneity of the remanence and the coercivity along the axial direction (Fig. 32). Moreover, the remanence of the same region of the RM also increased using the pre-deformation technique. The microstructure for regions 1, 2, 3 and 4 of the ring magnet confirms that the texture is improved in the regions shown, especially region 1 (Fig. 32). However, the alignment direction of the platelet Nd2Fe14B grains indicates a difference of almost 90° between region 1 and region 4, which could be ascribed to the rotation of the platelet Nd2Fe14B grains during BE. The Nd2Fe14B platelets in region 1 are inherited from the counterpart formed during the pre-deformation process and they rotate when they flow from the cylindrical part to the ring part (Fig. 31).

Fig. 32. (color online) (a) Magnetic performance and (b) the SEM images of the samples cut from different axial positions of the ring magnet prepared by pre-deformation and BE method. The white arrows represent the radial direction.[6]

Apart from the improvement of the remanence homogeneity, the enhancement of the magnetic performance was another focus of the RM research. Yi et al.[4] have investigated the effect of Nd content on the remanence of RM and the Nd content of 13.34 at.% was optimized. Gutfleisch et al.[79] determined that the deformation temperature and deformation speed have a great influence on the coercivity with little effect on the remanence of the RM prepared from the HDDR powder. A large deformation rate is beneficial to the coercivity, however, the deformation temperature has an optimum value of 725 °C. The optimized deformation temperature for melt-spun powder was determined by Li et al. to be 800 °C for texture formation and 660 °C for coercivity.[81] Moreover, a deformation rate of 0.01 and deformation ratio of 70% were determined to be optimum for the best texture and remanence.

5.3. Solution for tangential crack in NdFeB RM

Radially oriented RM prepared by the backward extrusion method usually exhibits cracks along the tangential direction on the inside surface in the top region (Fig. 33). The region with cracks is typically cut off which leads to wastage of Nd–Fe–B. To solve this problem, Satto et al.[82] employed an extra punch to press the cracks along the axial direction until they were healed. Dirba et al.[83] utilized a similar measure: they applied an axial pressure (not lower than 0.5 MPa) to the top region of the RM during backward extrusion. However, neither of these investigations indicated the reason for the tangential cracks. Previously, it was reported that the cracks were related to the low content of the Nd-rich phase,[84] because this phase acts as a glue during material flow. Moreover, Yan et al.[6] observed the phenomenon that no tangential cracks occurred in the middle or bottom region of the RM (Fig. 34). To determine the reason, they performed a backward extrusion over three packed samples (Fig. 35). The interface between the middle and bottom samples exhibits a small change (Fig. 36(a)), which indicates that no material flow took place in the bottom sample. In other words, material flow only takes place at a limited depth below the punch during backward extrusion. The top and middle samples exhibit an unusual shape and the interface between them is special (Fig. 36(b)). Most of the top sample was extruded into the ring magnet. However, the bottom region of the top sample exhibits a convex edge. Although it remains part of the top sample at this stage, it contributes mainly to the ring part of the middle sample. The top surface of the middle sample is concave-shaped, indicating the outward flow of Nd–Fe–B from the center, to form the ring magnet. The convex bottom of the top sample can be ascribed to the slower material flow in this region, due to the friction at the end surface of the punch. Further investigation revealed that the arc-shaped zone I (Fig. 37), which is in contact with the punch, plays the key role in the elimination of tangential cracks. Guruswamy et al.[85] reported on the existence of zones I and II using finite element analysis. Yin et al.[74] confirmed that the difference in texture is attributable to the flowing path. Moreover, different shapes of punches were redesigned and used for the preparation of RM (Fig. 38). The result shows that the half-spherical shape is the best one and could effectively eliminate the cracks in the top region of the RM.

Fig. 33. Tangential cracks in the top region of ring magnet.[6]
Fig. 34. Image on the vertical section of the backward-extruded ring magnet.[6]
Fig. 35. (color online) Schematic illustration of backward extrusion of the stacked HP magnets which were trisected from a mother HP magnet.[6]
Fig. 36. Photograph and schematic illustration of the bottom (a), middle (b), and top (b) samples after backward extrusion of stacked trisected hot pressing precursor. The height of the ring-shaped part is 7.75 mm.[6]
Fig. 37. Illustration on the three zones in cylindrical metals or alloys that were freely deformed under uniaxial pressure.[6]
Fig. 38. Images of the different shapes of punches and the ring magnets prepared with them: (a) half-spherical-end punch; (b) arc-shaped-end punch; (c) beveling cylinder; (d) cylinder.[6]
6. Physical properties of HD Nd–Fe–B magnets
6.1. Mechanical properties of HD NdFeB magnet

Sintered Nd–Fe–B magnets were initially reported to show an inferior fracture toughness compared to metals.[86] Their fracture toughness (K IC) is approximately 1–2 magnitudes lower than that of metals.[87,88] This fragile characteristic shortens the life of devices and is undesirable for machinery. HD Nd–Fe–B magnets have the characteristics of both a laminated structure consisting of Nd2Fe14B platelets and anisotropic grain boundaries (GB) (Figs. 39 and 40). This structure, which is different from that of a traditional sintered Nd–Fe–B magnet, could lead to inferior mechanical properties in the case of HD magnets. Jin et al.[88] investigated the mechanical properties of HD Nd–Fe–B magnets and revealed a significant anisotropy of the mechanical properties (Fig. 41). The compression strength, bending strength, and fracture toughness along the direction vertical to the Nd2Fe14B platelets are all higher than that in the direction parallel to the platelets. The origin of this phenomenon can be explained by the laminated structure of the Nd2Fe14B platelets (Fig. 42). In the case of HD Nd–Fe–B magnets, the fracture toughness is an important parameter in electric motor application and machinery. The toughness (K IC) along the direction vertical to Nd2Fe14B platelets is , while the value in the parallel direction to the platelets is only (Table 1). In fact, the sub-optimal properties in machinery or applications for HD magnets are attributed to the inferior mechanical properties along the direction parallel to the platelets. Therefore, it is important to improve the toughness in this direction and/or eliminate the anisotropy of toughness.

Fig. 39. Microstructure of hot-deformed (a) and sintered (b) Nd–Fe–B magnets.
Fig. 40. (color online) Schematic illustration of the microstructure of hot-deformed (a) and sintered (b) Nd–Fe–B magnets.
Fig. 41. Compression strength (a), bending strength (b), and fracture toughness (c) of HD Nd–Fe–B magnets along the directions that are parallel (dot-slash line) and vertical (straight line) to Nd2Fe14B platelets.[89]
Fig. 42. (color online) Sketch map of crack propagation along the grain boundary in HD magnets.[89]
Table 1.

Compression strength (σ c), bending strength (σ b), and fracture toughness (K IC) of HD Nd–Fe–B magnet along the directions parallel ( ) and vertical ( ) to Nd2Fe14B platelets. The deformation ratio of HD Nd–Fe–B magnet is 70%.

.

Since HD Nd–Fe–B magnets are easily cracked or broken along the GB parallel to the Nd2Fe14B platelets (GB-P), there are two methods that can be implemented to improve the toughness in this direction: the accumulative precipitation of nano-sized ductile phase and the dispersion of hard metal particles in GB-P (Fig. 43). Both approaches inhibit crack propagation or change the path of crack propagation in GB-P, which could improve the mechanical performance along the direction parallel to the Nd2Fe14B platelets in HD Nd–Fe–B magnets. Although such investigations in HD magnets have rarely been reported,[9092] similar studies have been performed on sintered Nd–Fe–B magnets. Wang et al.[93] have reported on the enhancement of impact toughness by Dy addition in sintered Nd–Fe–B magnets, while the negative effect by Pr addition. Li[94] demonstrated the positive effect of trace B addition on the bending strength. Other metals and alloys such as Ti, Co, Ni, M 38Co46.4Ni15.5 alloy (M is Ce-based alloy), Nd, etc. have also been added to Nd–Fe–B magnets to improve their mechanical properties.[9597] A systematic investigation on the effect of metal addition (Al, Ga, Cu, Nb, Zr, Ti, V, Mn, etc.) on the impact toughness has been performed by Liu et al.[98] Four ideal metals (Al, Ga, Cu, Nb) were shown to be efficacious in improving impact toughness. An examination of the mechanism indicated the in-situ precipitation of nano-sized ductile phase in GB after metal addition. The same mechanism was reported by Zeng et al.,[99] where Nd–(Fe, Co)–Cu and rod-like Nb–Fe–B phases were precipitated, and improved the bending strength of Nd–Fe–B magnets after the addition of Cu and Nb, respectively.

Fig. 43. (color online) Illustration of the two proposed methods for the enhancement of toughness. (a) Precipitation of nano-sized ductile phase and (b) dispersion of metal particles in the grain boundary parallel to Nd2Fe14B platelets.

In addition, other preparation parameters have been optimized to improve mechanical performance. Yi et al.[17] improved the bending strength of HD magnets by the addition of Cu powder. Ju et al.[100] tested the bending strength and toughness of these magnets using different Nd content, and discovered that the bending strength perpendicular to the Nd2Fe14B platelets increased with the increment of RE content, while a corresponding decrease was observed parallel to Nd2Fe14B platelets. Lin et al.[16] revealed the coarse grains could reduce both the compressing strength and the bending strength of HD magnets. This result is consistent with the finding that ultrafine grain is effective for improving impact toughness and bending strength.[101] Hu et al.[102,103] studied the mechanical performance of the HD magnet for different deformation rates. It was determined that both the Vickers hardness and the fracture toughness reached maximum values at a deformation rate of 40%. Hu et al.[104] also optimized the HD temperature for the Rockwell hardness and fracture toughness. Recently, Zheng et al.[105] determined that the cracks of hot-deformed Nd–Fe–B magnets are located in the ribbons boundaries comprised of non-oriented coarse grains. By suppressing the interfacial coarse grains, the mechanical properties can be effectively improved. This improvement can be ascribed to the refined grain size and reduced stress concentration.

6.2. Electrical properties of HD NdFeB magnet

Nd–Fe–B magnets can be utilized in the rotators of brushless motors to achieve large torques and high efficiency. Nevertheless, these magnets are good conductors and eddy current loss during rotation in an alternating electromagnetic field is expected. The eddy current cause an increase in the operating temperature of the magnet which reduces the remanence and torque. As such, suppression of the increase in temperature during the application process is favorable to the improvement of magnet stability.

One commonly used approach is to increase the electric resistivity of Nd–Fe–B magnets. Since the grain boundary exhibits a much lower resistivity than the Nd2Fe14B phase, increasing the resistivity of the grain boundary is key. In this regard, Marinescu et al.[106] have investigated the use of fluorides including NdF3, DyF3, and CaF2. The results show that CaF2 is the most stable additive after hot deformation and brings the highest resistivity for Pr-Fe-B HD magnets. Furthermore, the coercivity of the HD magnet remained nearly constant. Zheng et al.[107,108] introduced Al2O3 and SiO2 nanoparticles into Nd–Fe–B HD magnets. Both of these nanoparticles are able to increase the resistivity, but the remanence and coercivity were observed to significantly decrease. The deterioration of magnetic performance was due to the introduction of the element oxygen and atom diffusion during the HD process. After that, Wang et al.[109] has developed a facile method in which NdF3 powder was added into Nd–Fe–B magnet in a stacking way (Fig. 44). Electric resistivity increased monotonically with the increase of NdF3 addition, due to the formation of continuous segmented fluoride layers. The electric resistivity increases by nearly eight times when NdF3 content reached 5.3 wt.% (Fig. 45).

Fig. 44. (color online) Back scattered electron SEM image for Nd2Fe14B/NdF3 (2.12 wt.%) hot-deformed magnets with laminated structure and the element concentration maps for Fe, Nd, and F. Arrow is the pressure direction.[109]
Fig. 45. Dependence of the electrical resistivity of Nd2Fe14B/NdF3 on the percentage of NdF3.[109]

Subsequently, Zheng[110,111] reported similar works and prepared a Nd–Fe–B/CaF2 composite magnet with which the effect of CaF2 addition on the properties of the HD Nd–Fe–B magnet was investigated. Interestingly, when the mixtures were blended in ethanol, CaF2 was observed to transform from discontinuous blocks to continuous laminated structures in the composite magnet. The resistivity of the composite magnet increased with the increment of CaF2 and reached a maximum value of when the content of CaF2 reached 20 vol.%. However, the maximum energy product was reduced.

In addition, Sawatzki et al.[112] also enhanced the electrical resistivity of Nd–Fe–B hot-deformed magnets via adding DyF3 powders. The DyF3 additive that is enriched at the particle boundaries forms insulating Dy–F rich inclusions, which improve the electrical resistivity of local regions (Fig. 46).

Fig. 46. BSE image on a polished cross-section shows the electrically insulating Dy–F rich inclusions (bright area) within the MQU-F flakes (dark areas). The white arrows indicate the pressing direction.[112]
Reference
[1] Lee R W 1985 Appl. Phys. Lett. 46 790
[2] Luo Yang Zhang N 1989 Proc. 10th International workshop on Rare Earth Magnets and Their Application 1989 Kyoto, Japan 275
[3] Mishra R K 1987 J. Appl. Phys. 62 967
[4] Yi P P Lee D Yan A R 2010 J. Magn. Magn. Mater. 322 3019
[5] Yin W Z Chen R J Tang X Fu W B Lee D Yan A R 2014 Int. Conf. NdFeB Magnets: Supply Chain Crit. Properties &Appl. March 3-5 2014 Ningbo, China 258
[6] Yan A R Zhang C 2014 New-type Rare Earh Permanent Magn. Mater. Permanent Magent Motors Beijing Sci. Press 194 (in Chinese)
[7] Croat J J 1989 IEEE Trans. Magn. 25 3550
[8] Tang X Sepehri-Amin H Ohkubo T Hioki K Hattori A Hono K 2017 Acta Mate. 123 1
[9] Fuerst C D Brewer E G 1993 J. Appl. Phys. 73 5751
[10] Jin C X Chen R J Yin W Z Tang X Wang Z X Lee D Yan A R 2014 IEEE Trans. Magn. 50 1
[11] Lipiec W Davies H A 2010 J. Alloys. Compd. 491 694
[12] Li W Wang H J Lin M Lai B Li D Pan W 2011 J. Magn. 16 300
[13] Liu J Sepehri-Amin H Ohkubo T Hioki K Hattori A Schrefl T Hono K 2013 Acta Mater. 61 5387
[14] Yi P P Lin M Chen R J Lee D Yan A R 2010 J. Alloys Compd. 491 605
[15] Li L Graham C D 1990 J. Appl. Phys. 67 4756
[16] Lin M Wang H J Yi P P Yan A R 2010 J. Magn. Magn. Mater. 322 2268
[17] Zhao R Zhang W C Li J J Wang H J Zhu M G Li W 2011 J. Magn. 16 294
[18] Hu Z H Chu L H Li J Liu Y 2011 J. Rare Earths. 29 660
[19] Liu W Q Cui Z Z Yi X F Yue M Jiang Y B Zhang D T Zhang J X Liu X B 2010 J. Appl. Phys. 107 09A719
[20] Grünberger W Hinz D Kirchner A Müller K H Schultz L 1997 J. Alloys Compd. 257 293
[21] Hou Y H Huang Y L Liu Z W Zeng D C Ma S C Zhong Z C 2013 Mater. Sci. Eng. B 178 990
[22] Lai B Li Y F Wang H J Li A H Zhu M G 2013 J. Rare. Earths 31 679
[23] Yin W Z Chen R J Tang X Tang X Lee D Yan A R 2014 IEEE Trans. Magn. 50 1
[24] Liu Y H Guo S Liu X M Lee D Yan A R 2012 J. Appl. Phys. 111 07A705
[25] Cao X J Chen L Guo S Chen R J Yan G L Yan A R 2016 Scr. Mater. 116 40
[26] Tang X Chen R J Yin W Z Lin M Lee D Yan A R 2012 J. Appl. Phys. 111 07B540
[27] Ma Y L Chen D M Zhou A R Sun J C Cao P J 2012 Physica B 407 4562
[28] Li Y Kim Y B Yoon T S Jin D Z Suhr D S Kim C O 2002 J. Appl. Phys. 91 7887
[29] Liu Y Ma Y L Li J Li C Y Xie F Y Chu L H 2010 IEEE Trans. Magn. 46 2566
[30] Saito T Fujita M Kuji T Fukuoka K Syono Y 1998 J. Appl. Phys. 83 6390
[31] Yoshikawa N Kasai Y Watanabe T Shibata S Panchanathan V Croat J J 1991 J. Appl. Phys. 69 6049
[32] Yue M Zuo J H Liu W Q Lv W C Zhang D T Zhang J X Guo Z H Li W 2011 J. Appl. Phys. 109 07A711
[33] Ma Q Yue M Lv W C Zhang H G Yuan X K Zhang D T Zhang X F Zhang J X Gao X X 2016 J. Magn. 21 25
[34] De Campos M F Sampaio da Silva F A Périgo E A Castro J A D 2013 J. Magn. Magn. Mater. 345 147
[35] Pinkerton F E Fuerst C D 1991 J. Appl. Phys. 69 5817
[36] Volkov V V Zhu Y 1999 J. Appl. Phys. 85 3254
[37] Fuerst C D Brewer E G 1990 Appl. Phys. Lett. 56 2252
[38] Fuerst C D Brewer E G 1991 J. Appl. Phys. 69 5826
[39] Tang X Chen R J Yin W Z Lin M Lee D Yan A R 2013 IEEE Trans. Magn. 49 3237
[40] Sawatzki S Dirba I Wendrock H Schultz L Gutfleisch O 2014 J. Magn. Magn. Mater. 358 163
[41] Zheng X F Li M Chen R J Lei F Jin C X Wang Z X Ju J Y Yin W Z Lee D Yan A R 2017 Scr. Mater. 132 49
[42] Sepehri-Amin H Ohkubo T Nishiu chi T Hirosawa S Hono K 2010 Scr. Mater. 63 1124
[43] Sepehri-Amin H Prabhu D Hayashi M Ohkubo T Hioki K Hattori A Hono K 2013 Scr. Mater. 68 167
[44] Sepehri-Amin H Ohkubo T Nagashima S Shoji T Kato A Schrefl T Hono K 2013 Acta Mater. 61 6622
[45] Sepehri-Amin H Liu L Ohkubo T Yano M Shoji T Kato A Hono K 2013 Scr. Mater. 69 647
[46] Sepehri-Amin H Liu L H Ohkubo T Yano M Shoji T Kato A Schrefl T Hono K 2015 Acta Mater. 99 297
[47] Akiya T Liu J Sepehri-Amin H Ohkubo T Hioki K Hattori A Hono K 2014 J. Appl. Phys. 115 17A766
[48] Tang X Chen R J Yin W Z Jin C X Lee D Yan A R 2015 Appl. Phys. Lett. 107 202403
[49] Seelam U M R Liu L H Akiya T Sepehri-Amin H Ohkubo T Sakuma N Yano M Kato A Hono K 2016 J. Magn. Magn. Mater. 412 234
[50] Bance S Oezelt H Schrefl T Ciuta G Dempsey N M Givord D Winklhofer M Hrkac G Zimanyi G Gutfleisch O Woodcock T G Shoji T Yano M Kato A Manabe A 2014 Appl. Phys. Lett. 104 182408
[51] Akiya T Liu J Sepehri-Amin H Ohkubo T Hioki K Hattori A Hono K 2014 Scr. Mater. 81 48
[52] Liu L H Sepehri-Amin H Ohkubo T Yano M Kato A Shoji T Hono K 2017 Scr. Mater. 129 44
[53] Wang Z X Ju J Y Wang J Z Yin W Z Chen R J Li M Jin C X Tang X Don Lee Yan A R 2016 Sci. Rep. 6 38335
[54] Cui W B Takahashi Y K Hono K 2011 Acta Mater. 59 7768
[55] Dempsey N M Woodcock T G Sepehri-AminSepehri-Amin Zhang Y Kennedy H Givord D Hono K Gutfleisch O 2013 Acta Mater. 61 4920
[56] Liu J P 2009 Exchange-Coupled Nanocomposite Permanent Magnets New York Springer 309
[57] Kwon H W Hadjipanayis G C 2005 IEEE Trans. Magn. 41 3856
[58] Gabay A M Zhang Y Hadjipanayis G C 2004 Appl. Phys. Lett. 85 446
[59] Lee D Bauser S Higgins A Chen C Liu S 2006 J. Appl. Phys. 99 08B516
[60] Li L Graham C D 1992 IEEE Trans. Magn. 28 2130
[61] Ohmori K Li L Graham C D 1992 IEEE Trans. Magn. 28 2139
[62] Lee D Hilton J S Liu S Zhang Y J Hadjipanayis G C Chen C H 2003 IEEE Trans. Magn. 39 2947
[63] Liu Y G Xu L Guo D F Li W Wang Q F Zhang X Y 2009 J. Appl. Phys. 106 113918
[64] Liu Y G Xu L Guo D F Li W Wang Q F Zhang X Y 2009 Appl. Phys. Lett. 94 172502
[65] Li W Li X H Guo D F Sato K Gunderov D Stolyarov V V Zhang X Y 2009 Appl. Phys. Lett. 94 231904
[66] Kwon H W Yu J H 2009 IEEE Trans. Magn. 45 4435
[67] Li W Li L L Nan Y Li X H Zhang X Y Gunderov D Stolyarov V V Popov A 2007 Appl. Phys. Lett. 91 062509
[68] Liu S Higgins A Shin E Bauser S Chen C Lee D Shen Y He Y Huang M Q 2006 IEEE Trans. Magn. 42 2912
[69] Tang X Chen R J Yin W Z Wang J Z Tang X Lee D Yan A R 2014 Scr. Mater. 88 49
[70] Tang X Chen R J Yin W Z Wang J Z Tang X Lee D Yan A R 2013 Appl. Phys. Lett. 102 072409
[71] Hinz D Kirchner A Brown D N Ma B M Gutfleisch O 2003 J. Mater. Proc. Technol. 135 358
[72] Li A H Li W Lai B Wang H J Zhu M G Pan W 2010 J. Appl. Phys. 107 09A725
[73] Kim H T Kim Y B 2004 Phys. Status Solidi A 201 1938
[74] Yin W Z Chen R J Tang X Lin M Lee D Yan A R 2012 J. Appl. Phys. 111 07A727
[75] Yuri T Ohki T 1994 Proc. 13th International Workshop on Rare Earth Magnets and their Applications 1994 United Kingdom Birmingham
[76] Leonowicz M Derewnicka D Wozniak Davies H A 2004 J. Mater. Proc. Technol. 153 860
[77] Grunberger W Hinz D Schlafer D Schultz L 1996 J. Magn. Magn. Mater. 157 41
[78] Gutfleisch O Kirchner A Grunberger W Hinz D Nagel H Thompson P Chapman J N Muller K H Schultz L Harris I R 1998 J. Phys. D: Appl. Phys. 31 807
[79] Gutfleisch O Kirchner A Grunberger W Hinz D Schafer R Schultz L Harris I R 1998 J. Magn. Magn. Mater. 183 359
[80] X. Tang, R.J. Chen, W.Z. Yin, M. Lin, D. Lee, A.R. Yan, (China Patent) Z L 201110371635.9 [2015-09-09].
[81] Li A H Zhao R Lai B Wang H J Zhu M G Li W 2011 Chin. Phys. B 20 107503
[82] M. Saito, T. Watanabe, S. Yahagi, Y. Kasai, N. Yoshikawa, Y. Yoshida, T. Kinami, H. Yamada, (European Patent) EP0392799 (A1) [1990-10-17]
[83] Dirba I Sawatzki S Gutfleisch O 2014 J. Alloys Compd. 589 301
[84] Brown D N Smith B Ma B M Campbell P 2004 IEEE Trans. Magn. 40 2895
[85] Guruswamy S Wang Y R Panchanathan V 1998 J. Appl. Phys. 83 6393
[86] Wang Q F Zhang M G 2010 Shanxi Metallurgy 33 11 (in Chinese)
[87] Li A H Dong S Z Li W 2002 Met. Func. Mater. 9 7 (in Chinese)
[88] Li W Li A H Wang H J 2005 IEEE Trans. Magn. 41 2339
[89] Jin C X Chen R J Yin W Z Tang X Wang Z X Lee D Yan A R 2014 IEEE Trans. Magn. 50 2103204
[90] Hu Z H Dong H Ma D W Luo C 2016 J. Magn. Magn. Mater. 401 169
[91] Lin M Wang H J Zheng J W Yan A R 2015 J. Magn. Magn. Mater. 379 90
[92] Horton J A Wright J L Herchenroeder J W 1996 IEEE Trans. Magn. 32 4374
[93] Wang H J Li A H Li W 2007 Intermetallics 15 985
[94] Li A H Li W Dong S Z Li X M 2003 J. Rare Met. 27 531
[95] Shimotomai M Fukuda Y Fujita A Ozak Y Kitano Y 1994 Bull. Jpn. Inst. Met. 33 619
[96] Hosseini H R M Kianvash A 2006 Mater. Lett. 4 555
[97] Wang H J Li A H Zhu M G Li W 2006 J. Magn. Magn. Mater. 307 268
[98] Liu J F Vora P Walmer M H Kottcamp E Bauser S A Higgins A Liu S 2005 J. Appl. Phys. 97 10H101
[99] Zeng Z P Jiang J H 1999 Part. A: Phys. Test. 3 202
[100] Ju J Y Tang X Chen R J Yan A R Jin C X Yin W Z Wang Z X Lee D Zhang Z M 2015 J. Magn. Magn. Mater. 386 31
[101] Yue M Cao A L Wang G P Liu W Q Zhang D T Zhang J X 2008 J. Func. Mate. Dev. 14 682
[102] Hu Z H Qu H J Zhao J Q Luo C Li J Liu Y 2012 J. Rare. Earths 30 1112
[103] Hu Z H Chu L H Li J Liu Y 2012 J. Magn. Magn. Mater. 324 101
[104] Hu Z H Li J Chu L H Liu Y 2011 J. Magn. Magn. Mater. 323 104
[105] Zheng X F Li M Jin C X Chen R J Yin W Z Tang X Lei F Wang Z X Ju J Y Lee D Yan A R 2017 J. Alloys Compd. 728 607
[106] Marinescu M Gabay A M Liu J F Hadjipanayis G C 2009 J. Appl. Phys. 105 07A711
[107] Zheng L Y Li W Zhu M G Ye L P Bi W C 2013 J. Alloys Compd. 560 80
[108] Zheng L Y Ye L P Bi W C Li W Zhu M G 2013 J. Magn. Mater. Dev. 44 27
[109] Wang Z X Chen R J Yin W Z Yan C J Tang X Jin C X Lee D Yan A R 2014 IEEE Trans. Magn. 50 2104103
[110] Li W Zheng L Y Zhu M G Xin H H Zheng D W 2014 IEEE Trans. Magn. 50 2103303
[111] Zheng L Y Li W Zhu M G Xin H H Zheng D W Zhao L X 2015 IEEE Trans. Magn. 51 2102204
[112] Sawatzki S Dirba I Schultz L Gutfleisch O 2013 J. Appl. Phys. 114 133902